Comparative study on HSI classification with  GF and RF

Ms. Vidya R Menon; Mr. Shibi Thambi k; Mr. Sreerag. S

CALL FOR PAPERS : Feb-2026

Submission Last Date

25-Feb-26

Submit Manuscript Online

FOR AUTHORS

FOR REVIEWERS

ARCHIEVES

DOWNLOADS

Open Access

Comparative study on HSI classification with GF and RF

Author(s):

Ms. Vidya R Menon , Thejus Engg College ; Mr. Shibi Thambi k , Thejus Engg College; Mr. Sreerag. S , Thejus Engg College

Keywords:

Feature extraction, guided filtering, hyperspectral image, image classification, image fusion (IF), recursive filtering

Abstract:

In the area of remote sensing as well as the geosciences, whenever we are capturing a particular scene, it becomes very difficult to identify each and every object in the image of a particular scene. Here comes the advantage of using hyperspectral imaging as well as the feature extraction method. In hyper spectral imaging, the recorded spectra have fine wavelength resolution and cover a wide range of wavelengths. The goal of hyper spectral imaging is to obtain the spectrum for each pixel in the image of a scene. And it is done with the purpose of finding objects, identifying materials or detecting processes. Feature extraction is an effective way in order to reduce computational complexity and to increase the accuracy of hyper spectral image classification. Here, a comparative study on two simple and efficient feature extraction methods has been proposed. The work is based on image fusion and two different filtering approaches i.e.; recursive filtering and guided filtering. First, the partitioning of hyper spectral image into multiple subsets of adjacent hyper spectral bands was done. Then, the bands in each subset are getting fused together by a simplest image fusion method called averaging. After that, the fused bands are processed with transform domain recursive filtering as well as guided filtering in order to get the resulting features for classification. Finally, we will get a classification map which will indicate each and every object in the image of a particular scene. The performance of these two methods has been compared using various performance measures.